Requity distribution method, device, equipment, medium and program product

Abstract

The embodiment of the invention discloses a right distribution method, device and equipment, a medium and a program product. Comprising the steps of obtaining a current marketing activity category, a current user portrait category and user behavior data; inputting the current marketing activity category, the current user portrait category and the user behavior data into a pre-trained traffic prediction model, and determining a traffic prediction value; and adjusting attribute information of the dynamic token bucket according to the flow prediction value, and performing right distribution through the dynamic token bucket. The data collected from multiple angles or levels is used as the input of the traffic prediction model, so that the user demand and the market environment can be reflected more comprehensively, and the accuracy of the determined traffic prediction value is improved. The resource in the dynamic token bucket can be dynamically adjusted according to the traffic demand, so that the matching degree of the resource provided by the token bucket and the resource required for right distribution is improved, insufficient resource distribution and resource waste caused by customer access traffic fluctuation are avoided, and the right can be smoothly distributed.

Description

Technical Field

[0001] The present invention relates to the field of traffic management technology, and in particular to a rights distribution method, device, equipment, medium and program product. Background Art

[0002] With the rapid development of mobile Internet, the number of mobile banking users continues to grow and marketing activities become increasingly frequent. As an important means to attract users, the stability and efficiency of rights distribution are particularly important.

[0003] In traditional mobile banking marketing activities, token bucket flow control is usually used to control the speed of rights distribution. However, due to fluctuations in customer access traffic, this method often results in token buckets not being fully utilized during low traffic periods, and being quickly exhausted during peak traffic periods. This unstable distribution speed not only affects user experience, but can also lead to resource waste or a situation where supply exceeds demand.

[0004] Currently, various problems caused by fluctuations in customer access traffic are often solved by setting static token bucket current limits, increasing server resources, and optimizing network architecture. However, due to the volatility of traffic, the setting of static token bucket current limits lacks flexibility and is difficult to dynamically adjust according to real-time traffic changes; increasing server resources can improve the processing capacity of the system, but the cost is high and it cannot fundamentally solve the problems caused by traffic fluctuations; although optimizing the network architecture can improve the stability and performance of the system, the optimization of the network architecture often requires a lot of time and effort, and the final optimization effect is difficult to guarantee. It can be considered that even if the system is optimized in the above way, it is difficult to meet the needs of high-utilization rights distribution in the case of fluctuations in customer access traffic. Summary of the invention

[0005] The present invention provides a rights distribution method, device, equipment, medium and program product. Based on various influencing factors that may cause fluctuations in customer access traffic in marketing activities, the traffic required for distributing rights to users during marketing activities is predicted, and a token bucket is dynamically adjusted based on the prediction result, so that the matching degree between the resources provided by the token bucket and the resources required for rights distribution is improved, avoiding insufficient resource allocation and resource waste due to fluctuations in customer access traffic, so that rights can be distributed smoothly.

[0006] In a first aspect, an embodiment of the present invention provides a method for allocating rights and interests, including:

[0007] Get the current marketing campaign category, current user profile category, and user behavior data;

[0008] Input the current marketing activity category, current user profile category and user behavior data into the pre-trained traffic prediction model to determine the traffic prediction value;

[0009] The attribute information of the dynamic token bucket is adjusted according to the traffic prediction value, and the rights and interests are distributed through the dynamic token bucket.

[0010] In a second aspect, an embodiment of the present invention further provides a rights distribution device, including:

[0011] Data acquisition module, used to obtain the current marketing activity category, current user portrait category and user behavior data;

[0012] The traffic prediction module is used to input the current marketing activity category, the current user profile category and the user behavior data into the pre-trained traffic prediction model to determine the traffic prediction value;

[0013] The equity distribution module is used to adjust the attribute information of the dynamic token bucket according to the traffic prediction value and distribute the equity through the dynamic token bucket.

[0014] In a third aspect, an embodiment of the present invention further provides a rights distribution device, the rights distribution device comprising:

[0015] at least one processor; and a memory communicatively coupled to the at least one processor;

[0016] The memory stores a computer program that can be executed by at least one processor, and the computer program is executed by at least one processor so that the at least one processor can implement the equity distribution method of any embodiment of the present invention.

[0017] In a fourth aspect, an embodiment of the present invention further provides a storage medium comprising computer executable instructions, which, when executed by a computer processor, are used to execute the equity distribution method of any embodiment of the present invention.

[0018] In a fifth aspect, an embodiment of the present invention further provides a computer program product, including a computer program, which, when executed by a processor, is used to execute the equity distribution method of any embodiment of the present invention.

[0019] The embodiment of the present invention provides a method, device, equipment, medium and program product for distributing rights and interests, by obtaining the current marketing activity category, the current user portrait category and the user behavior data; inputting the current marketing activity category, the current user portrait category and the user behavior data into the pre-trained traffic prediction model to determine the traffic prediction value; adjusting the attribute information of the dynamic token bucket according to the traffic prediction value, and distributing rights and interests through the dynamic token bucket. By adopting the above technical solution, the data collected from multiple angles or levels including the current marketing activity category, the current user portrait category and the user behavior data are used as the input of the traffic prediction model, and the traffic prediction value that the user of the user portrait category may need to distribute rights and interests under the current marketing activity is obtained. Since the data angles input into the traffic prediction model are diverse, it can more comprehensively reflect the user needs and the market environment, and improve the accuracy of the determined traffic prediction value. Then, the attribute information of the dynamic token bucket is adjusted according to the predicted traffic prediction value, so that the resources in the dynamic token bucket can be dynamically adjusted according to the traffic demand, so that the matching degree between the resources provided by the token bucket and the resources required for the rights and interests distribution is improved, and the insufficient resource allocation and resource waste caused by the fluctuation of the customer access traffic are avoided, so that the rights and interests can be distributed smoothly.

[0020] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0022] Figure 1 A schematic diagram of a flow chart of a method for allocating rights and interests provided in the first embodiment of the present invention;

[0023] Figure 2 A flowchart of a method for allocating rights and interests provided in Embodiment 2 of the present invention;

[0024] Figure 3 An example diagram of the training process of a multi-factor prediction model using both LSTM units and RNN units provided in the second embodiment of the present invention;

[0025] Figure 4 A schematic diagram of the structure of a rights distribution device provided in Embodiment 3 of the present invention;

[0026] Figure 5A schematic diagram of the structure of a rights distribution device provided in Embodiment 4 of the present invention. DETAILED DESCRIPTION

[0027] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0028] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0029] Embodiment 1

[0030] Figure 1 The flowchart of a method for distributing rights and interests provided in the first embodiment of the present invention is applicable to the situation where the distribution of rights and interests is controlled according to the customer access flow in the mobile banking marketing activities. The method can be executed by a rights and interests distribution device, which can be implemented by software and / or hardware, and the rights and interests distribution device can be configured in the rights and interests distribution device. Optionally, the rights and interests distribution device can be an electronic device, which can be a notebook, a desktop computer, a smart tablet, a mobile terminal, etc., and the embodiment of the present invention does not limit this.

[0031] like Figure 1 As shown, a method for distributing rights and interests provided by an embodiment of the present invention specifically includes the following steps:

[0032] S101. Obtain the current marketing activity category, current user portrait category, and user behavior data.

[0033] In this embodiment, the current marketing activity category can be specifically understood as the category to which the marketing activity conducted by mobile banking at the current moment belongs. It can be understood that the marketing activity category can be determined according to the characteristics of the marketing activity. Each marketing activity has its own unique objectives, audiences, time, location, and preferential methods. The marketing activities can be divided into multiple categories based on the characteristics of the activities. Different marketing activity categories have different attractiveness and influence on traffic. The division method of marketing activity categories in the embodiment of the present invention is not specifically limited. The current user portrait can be specifically understood as information that comprehensively describes the user characteristics of the user group for the marketing activities of the current marketing activity category. It can be understood that a user portrait information includes but is not limited to information such as age, gender, region, occupation, income level and consumption habits. The above information can help understand the composition and preferences of the user group, so as to more accurately determine the attractiveness of different marketing activities to them. User behavior data can be specifically understood as relevant data collected by the mobile terminal used by the user and the user's operations on the mobile terminal. Exemplarily, user behavior data may include data on the user's clicks, browsing, searches and purchases on the platform. The above information can be used to gain insight into the user's real needs and interest preferences.

[0034] Specifically, when it is necessary to carry out marketing activities, it is necessary to first estimate the customer traffic that can be brought by the marketing activities, so that sufficient resources can be reserved for user rights distribution after the marketing activities are carried out. Therefore, the current marketing activity category can be determined first according to the marketing activities to be carried out, and then based on the user groups that the current marketing activity category mainly targets, the current user portrait category corresponding to the current marketing activity category and the user behavior data generated by the interaction between users and the platform under the current user portrait category can be obtained.

[0035] S102: Input the current marketing activity category, the current user portrait category and the user behavior data into the pre-trained traffic prediction model to determine the traffic prediction value.

[0036] In this embodiment, the traffic prediction model can be specifically understood as a neural network model that is used to predict the traffic required for the equity method corresponding to the marketing activities and users based on the input of multiple types of information related to marketing activities and users, combining long-term trends with short-term fluctuations.

[0037] Specifically, the current marketing activity category, the current user portrait category and the user behavior data are input as the input parameters of the pre-trained traffic prediction model. The traffic prediction model extracts features of the input current marketing activity category, the current user portrait category and the user behavior data, and captures the long-term dependency between the extracted features and the long-term trends and cyclical patterns in the historical traffic data. Meanwhile, the changes in short-term fluctuations in marketing activities and user behaviors are captured, and the two are combined to output the traffic value that may be generated by the marketing activities to be carried out, and the traffic value is determined as the traffic prediction value.

[0038] S103: Adjust the attribute information of the dynamic token bucket according to the traffic prediction value, and distribute rights and interests through the dynamic token bucket.

[0039] In this embodiment, the token bucket can be specifically understood as an algorithm for computer network traffic shaping and flow control, the core idea of ​​which is to control the rate of data traffic through a virtual "bucket" in which consumable tokens are stored, and each token represents a certain number of traffic transmission permissions. The dynamic token bucket can be specifically understood as a token bucket that allows the token generation rate and bucket capacity to be dynamically adjusted according to network conditions or other factors. The attribute information can be specifically understood as information used to indicate the dynamic token bucket capacity and token generation speed.

[0040] Specifically, based on the traffic prediction value, it is possible to determine the traffic prediction value that may be generated by the customer's demand for rights and interests when conducting a marketing activity of the same type as the current marketing activity. At this time, the maximum number of rights and interests that need to be provided at the same time, as well as the minimum speed for external rights and interests distribution and other information can be determined based on the traffic prediction value. Based on the above information, the attribute information in the dynamic token bucket that indicates the bucket capacity and the token generation speed can be adaptively adjusted so that the adjusted dynamic token bucket matches the resources corresponding to the traffic prediction value. After completing the configuration of the dynamic token bucket, tokens can be taken out of the dynamic token bucket according to user needs, so that each token represents a distributable right and the rights and interests are distributed to users who request the rights and interests.

[0041] The technical solution of this embodiment obtains the current marketing activity category, the current user portrait category and the user behavior data; inputs the current marketing activity category, the current user portrait category and the user behavior data into the pre-trained traffic prediction model to determine the traffic prediction value; adjusts the attribute information of the dynamic token bucket according to the traffic prediction value, and distributes rights through the dynamic token bucket. By adopting the above technical solution, the data collected from multiple angles or levels, including the current marketing activity category, the current user portrait category and the user behavior data, are used as the input of the traffic prediction model to obtain the traffic prediction value that the user of the user portrait category may need to issue rights under the current marketing activity. Due to the diverse data angles input into the traffic prediction model, it can more comprehensively reflect user needs and market environment, and improve the accuracy of the determined traffic prediction value. Then, according to the predicted traffic prediction value, the attribute information of the dynamic token bucket is adjusted, so that the resources in the dynamic token bucket can be dynamically adjusted according to the traffic demand, so that the matching degree between the resources provided by the token bucket and the resources required for the rights distribution is improved, and the insufficient resource allocation and resource waste caused by the fluctuation of customer access traffic are avoided, so that the rights can be distributed smoothly.

[0042] Embodiment 2

[0043] Figure 2 A flow chart of a rights distribution method provided in the second embodiment of the present invention. The embodiment of the present invention is further optimized on the basis of the above-mentioned optional technical solutions. Before the traffic prediction model is put into use, the historical marketing activity categories, the historical user portrait categories corresponding to the historical marketing activity categories, the historical user behavior data and the historical traffic data corresponding to the historical user portrait categories are first obtained, and the traffic data under different marketing activity categories are segmented to comprehensively construct training samples for training the traffic prediction model. During the training process, the traffic prediction model can be trained for traffic prediction after further segmentation according to the user portrait categories for each historical marketing activity category, and the user behavior data can reflect the frequency and intensity of interaction between the user and the platform, that is, it can be used to evaluate the impact of user activity on traffic. During the training process, the traffic data is combined with the activity-related data to more comprehensively understand the user behavior and improve the prediction accuracy of the trained traffic prediction model. Then, dynamic adjustment of the dynamic token bucket capacity and token generation speed is achieved based on the traffic prediction value, so that when a user's rights request is received, the token can be taken out of the dynamic token bucket in time to distribute rights to the user. At the same time, the dynamic token bucket will not occupy too many resources, causing resource waste, reducing the cost of rights distribution and improving the smoothness of rights distribution.

[0044] like Figure 2 As shown, a method for distributing rights and interests provided by an embodiment of the present invention specifically includes the following steps:

[0045] S201. Obtain at least one historical marketing activity category, a historical user portrait category corresponding to each historical marketing activity category, and historical user behavior data and historical traffic data corresponding to each historical user portrait category.

[0046] In this embodiment, the historical marketing activity category can be specifically understood as the category to which the marketing activities that mobile banking has conducted within a period of historical time belong. Exemplarily, the historical marketing activity category may include categories such as post-transaction lucky draws, star-rated areas, and popular activities, which are not limited in this embodiment of the present invention. The historical user portrait category can be specifically understood as the user portrait category corresponding to the marketing activity users who responded to the historical marketing activity category within the same period of historical time. Exemplarily, the historical user portrait category may include information such as age, gender, and consumption habits, which is not limited in this embodiment of the present invention. The historical user behavior data can be specifically understood as the behavior data generated by the user corresponding to the historical user portrait category within the same period of historical time. The historical traffic data can be specifically understood as the segmented traffic data generated by the user after the historical user behavior data in a historical user portrait category under a historical marketing activity category.

[0047] S202: According to the correspondence between each historical marketing activity category, each historical user portrait category, each historical user behavior data and each historical traffic data, respectively construct training samples with historical traffic data as labels.

[0048] Specifically, since the correspondence between different historical marketing activity categories, historical user portrait categories, historical user behavior data and historical traffic data can be clarified during the acquisition process, that is, one historical marketing activity category may include multiple historical user portrait categories, and one historical user portrait category may include multiple historical user behavior data and corresponding historical traffic data, at this time, based on the most detailed correspondence, with each historical traffic data as the most detailed dimension, multiple training samples labeled with historical traffic data can be constructed.

[0049] For example, under the historical marketing activity category A, historical user profile categories B, C and D can be subdivided. The users corresponding to the historical user profile categories B, C and D will generate historical user behavior data, and the historical traffic data LB, LC and LD generated after the historical user behavior data is generated can be determined; that is, there are corresponding relationships of AB-LB, AC-LC and AD-LD. Then, training samples corresponding to AB, AC and AD marketing activities and user profiles can be constructed with LB, LC and LD as labels, respectively.

[0050] In an embodiment of the present invention, since the constructed training samples are subdivided into categories, each training sample simultaneously contains three dimensional data that affect the size of traffic data, namely, marketing activity category, user portrait category and user behavior data. This enriches the data dimensions in the training samples, enables a more comprehensive understanding of user behavior in response to marketing activities, and enables the traffic prediction model trained based on the training samples to perform more accurate traffic prediction.

[0051] S203: Train the initial traffic prediction model through each training sample to generate a pre-trained traffic prediction model.

[0052] In this embodiment, the initial traffic prediction model can be specifically understood as a traffic prediction model in which the weights of the neural network layers included therein are not adjusted, and it is consistent with the architecture of the traffic prediction model after training.

[0053] Specifically, each training sample is input into the initial traffic prediction model, and the weights of each neural network layer in the parameter traffic prediction model are iteratively adjusted by combining forward propagation and back propagation until the preset convergence conditions or other exit conditions are met, and the training is considered completed. The trained traffic prediction model is used as the pre-trained traffic prediction model required in subsequent use.

[0054] It is understandable that when training the traffic prediction model, the performance of the model is optimized by adjusting the model parameters and hyperparameters such as the learning rate, and an appropriate loss function (such as the mean square error) is used to measure the accuracy of the model prediction, and the back propagation algorithm is used to update the model weights. In order to improve the generalization ability and stability of the model, a series of regularization techniques and model tuning strategies are also used, such as Dropout and early stopping. In addition, cross-validation techniques can also be used to evaluate the performance of the model on different data sets to ensure that the model has good robustness.

[0055] Optionally, the pre-trained traffic prediction model is a multi-factor prediction model that uses both Long Short-Term Memory (LSTM) units and Recurrent Neural Network (RNN) units.

[0056] Among them, LSTM units can effectively capture long-term trends and periodic patterns in historical traffic data, while RNN units are good at processing short-term dependencies in sequence data and are very effective in capturing short-term changes such as marketing activities or customer behavior. By combining LSTM units with RNN units, the traffic prediction model can consider both long-term trends and short-term fluctuations, thereby making more accurate predictions.

[0057] The architecture of the long short-term memory network unit and the recurrent neural network unit in the pre-trained traffic prediction model includes at least one of the following:

[0058] Hybrid architecture;

[0059] Input-output hybrid architecture;

[0060] Multi-layered architecture.

[0061] In this embodiment, the hybrid architecture can be specifically understood as in a network, some layers use LSTM units, while other layers use traditional RNN units. For example, the RNN layer can be used to capture short-term dependencies first, and then the LSTM layer can be used to process long-term dependencies.

[0062] In this embodiment, the input-output hybrid architecture can be specifically understood as in a network, some features can be input into LSTM units, while other features can be input into RNN units. Then, the outputs of these two models can be combined as a comprehensive prediction result.

[0063] In this embodiment, the multi-level architecture can be specifically understood as in a deep network, the bottom layer uses RNN units to process short-term sequence information, while the upper layer uses LSTM units to process longer-term information transmitted from the lower layer.

[0064] It is understandable that LSTM units and RNN units can receive the same input data, which is usually serialized traffic data, as well as other relevant features (such as marketing campaign categories, customer profile categories, and activity data). The output of the LSTM unit can be the hidden state of each time step in the sequence, which captures long-term dependencies; the output of the RNN unit can be the hidden state of each time step in the sequence, which captures short-term dependencies. When the traffic prediction model is output, the outputs of the LSTM unit and the RNN unit will be combined through methods such as weighted summation and splicing, and the final prediction result will be determined as the traffic prediction value.

[0065] For example, Figure 3 This is an example diagram of the training process of a multi-factor prediction model using both LSTM units and RNN units provided in the second embodiment of the present invention. Figure 3 As shown, the training of RNN units can be achieved in the following ways:

[0066] (1) Forward Propagation

[0067] 1) For a given input sequence (such as time series data), the RNN is expanded by time steps, and each time step receives the current input and the hidden state of the previous time step.

[0068] 2) The hidden state is updated through an activation function (such as tanh) and passed to the next time step.

[0069] 3) At the last time step or at all time steps, the RNN produces output, depending on the specific task (such as sequence classification, sequence generation, etc.).

[0070] (2) Loss calculation

[0071] 1) Use an appropriate loss function (such as mean squared error (MSE) for regression tasks and cross entropy loss for classification tasks) to calculate the difference between the model output and the true label.

[0072] 2) For sequence-to-sequence tasks, the loss may be calculated at each time step and summed or averaged.

[0073] (3) Back propagation through time

[0074] 1) Due to the recurrent nature of RNNs, the gradients need to be back-propagated through the entire sequence. This is called Back-Propagation Through Time (BPTT).

[0075] 2) In BPTT, the gradient of the loss function with respect to the model parameters is calculated, which involves calculating the gradient of the hidden state and output at each time step.

[0076] (4) Parameter update

[0077] Use an optimization algorithm (such as stochastic gradient descent SGD, Adam, etc.) to update the weights and biases of the RNN according to the calculated gradients.

[0078] (5) Regularization and tuning

[0079] 1) To prevent overfitting, regularization techniques such as L1 and L2 regularization can be applied.

[0080] 2) Use strategies such as early stopping and learning rate decay to further tune the model.

[0081] For example, Figure 3 As shown, the training of LSTM units can be achieved in the following way:

[0082] (1) Forward Propagation

[0083] 1) LSTM is also expanded by time steps, and each time step receives the current input and the hidden state and memory cell state of the previous time step.

[0084] 2) The input gate determines which new information is stored in the memory cell, the forget gate determines which information is discarded from the memory cell, and the output gate determines the output of the current time step.

[0085] 3) The memory cell state is updated after being processed by the forget and input gates and passed to the next time step.

[0086] (2) Loss calculation

[0087] Same as RNN, an appropriate loss function is used to calculate the difference between the model output and the true label.

[0088] (3) Back propagation through time

[0089] Since LSTM also has a cyclic structure, it is also necessary to calculate the gradient through BPTT; the gradient calculation of LSTM is more complicated than that of RNN because it involves the gating mechanism and the update of memory units.

[0090] (4) Parameter update

[0091] Use an optimization algorithm to update the weights and biases of the LSTM according to the calculated gradients.

[0092] (5) Regularization and Tuning

[0093] LSTM can also apply regularization techniques to prevent overfitting; the model can be further tuned by adjusting hyperparameters (such as the number of layers, number of hidden units, learning rate, etc.).

[0094] Understandably, when training LSTM and RNN, we also need to pay attention to the gradient vanishing and gradient exploding problems. These problems are particularly prominent when processing long sequences and can be alleviated by using specific initialization methods (such as He initialization), gradient clipping and other techniques.

[0095] S204: Obtain the current marketing activity category, current user portrait category, and user behavior data.

[0096] S205: Input the current marketing activity category, the current user portrait category and the user behavior data into the pre-trained traffic prediction model to determine the traffic prediction value.

[0097] S206: Determine attribute information of the dynamic token bucket according to the traffic prediction value and the pre-built traffic parameter correspondence.

[0098] The pre-built traffic parameter correspondence includes at least one of a ratio conversion relationship, a dictionary lookup mechanism based on business scale, and a rule engine;

[0099] Among them, the attribute information includes capacity and token generation speed;

[0100] The token generation speed is the speed at which tokens are generated and placed in the dynamic token bucket, and the capacity is the number of tokens that the dynamic token bucket can carry.

[0101] Specifically, the traffic prediction value is substituted into the pre-built traffic parameter correspondence relationship to determine the capacity and token generation speed required for the dynamic token bucket in order to provide sufficient traffic for user rights distribution without wasting resources when the customer traffic in the future is the traffic prediction value, and the determined capacity and token generation speed are determined as the attribute information of the dynamic token bucket.

[0102] Exemplarily, if the traffic parameter correspondence is a ratio conversion relationship, the traffic prediction value can be substituted into the ratio conversion relationship, and the traffic prediction value can be multiplied by the corresponding ratio to obtain the capacity of the dynamic token bucket and the token generation speed. If the traffic parameter correspondence is a dictionary lookup mechanism based on the business scale, the traffic prediction value can be used as a known quantity to look up in a pre-built dictionary to determine the capacity of the dynamic token bucket and the token generation speed. If the traffic parameter correspondence is a conversion relationship based on a rule engine, the traffic prediction value is substituted into the rule engine, and the capacity of the dynamic token bucket and the token generation speed are obtained by the output of the rule engine.

[0103] S207: When receiving a user rights request, take out a token from the dynamic token bucket and distribute rights to the user through the token.

[0104] Specifically, after the adjustment of the attribute information of the dynamic token bucket is completed, the dynamic token bucket will obtain tokens at the token generation speed until the capacity of the dynamic token bucket is reached. In this process, the user's rights and interests request given by the user for the current marketing activity can be received in real time, and each time a user's rights and interests request is received, a token is taken out from the dynamic token bucket, and the rights and interests are distributed to the user through the traffic transmission permission corresponding to the token.

[0105] Optionally, after the rights and interests are distributed through the dynamic token bucket, it also includes: obtaining user feedback data, and building supplementary training samples based on the current marketing activity category, the current user portrait category, user behavior data and user feedback data; and supplementing the training of the traffic prediction model through the supplementary training samples.

[0106] Specifically, in order to ensure that the processing method for rights distribution after going online can still be optimized, after the rights are distributed through the dynamic token bucket, the embodiment of the present invention will obtain user feedback data generated by users regarding information feedback such as whether the rights are distributed in a timely manner, and combine the user feedback data with the current marketing activity category, current user portrait category and user behavior data corresponding to the generation of the user feedback data, to form a supplementary training sample for supplementary training of the traffic prediction model, and the traffic prediction model can be supplemented by training the supplementary training samples, so that the traffic prediction model can continuously learn and improve during the rights distribution process, thereby enabling the entire distribution process to better adapt to market changes and changes in user needs.

[0107] The technical solution of this embodiment, before the traffic prediction model is put into use, firstly, by acquiring the historical marketing activity categories, the historical user portrait categories corresponding to the historical marketing activity categories, the historical user behavior data corresponding to the historical user portrait categories, and the historical traffic data, after the traffic data under different marketing activity categories are segmented, a training sample for training the traffic prediction model is comprehensively constructed, so that the traffic prediction model can be trained for each historical marketing activity category in the training process, and the traffic prediction after further segmentation according to the user portrait category, and the user behavior data can reflect the frequency and intensity of interaction between the user and the platform, that is, it can be used to evaluate the impact of user activity on traffic, and the traffic data is combined with the activity-related data during the training process, so that the user behavior can be more comprehensively understood, and the prediction accuracy of the traffic prediction model obtained by training is improved. Then, according to the traffic prediction value, the dynamic token bucket capacity and the token generation speed are dynamically adjusted, so that when a user's rights request is received, the token can be taken out of the dynamic token bucket in time to realize the rights distribution to the user, and at the same time, the dynamic token bucket will not occupy too many resources, resulting in resource waste, reducing the cost of rights distribution and improving the smoothness of rights distribution.

[0108] Embodiment 3

[0109] Figure 4 This is a schematic diagram of the structure of a rights distribution device provided in the third embodiment of the present invention, such as Figure 4 As shown, the equity distribution device includes a data acquisition module 31, a traffic prediction module 32 and an equity distribution module 33.

[0110] Among them, the data acquisition module 31 is used to obtain the current marketing activity category, the current user portrait category and the user behavior data; the traffic prediction module 32 is used to input the current marketing activity category, the current user portrait category and the user behavior data into the pre-trained traffic prediction model to determine the traffic prediction value; the equity distribution module 33 is used to adjust the attribute information of the dynamic token bucket according to the traffic prediction value, and distribute equity through the dynamic token bucket.

[0111] The technical solution of the embodiment of the present invention uses data collected from multiple angles or levels, including the current marketing activity category, the current user portrait category, and user behavior data, as the input of the traffic prediction model to obtain the traffic prediction value that users of the user portrait category may need to issue rights under the current marketing activity. Since the data angles of the input traffic prediction model are diverse, it can more comprehensively reflect user needs and market environment, and improve the accuracy of the determined traffic prediction value. Then, the attribute information of the dynamic token bucket is adjusted according to the predicted traffic prediction value, so that the resources in the dynamic token bucket can be dynamically adjusted according to the traffic demand, so that the matching degree between the resources provided by the token bucket and the resources required for the rights distribution is improved, avoiding insufficient resource allocation and resource waste due to fluctuations in customer access traffic, so that rights can be distributed smoothly.

[0112] Optionally, the equity distribution module 33 is specifically used for:

[0113] Determine the attribute information of the dynamic token bucket according to the traffic prediction value and the pre-built traffic parameter correspondence;

[0114] When receiving a user's rights and interests request, the token is taken out from the dynamic token bucket and the rights and interests are distributed to the user through the token;

[0115] The pre-built traffic parameter correspondence includes at least one of a ratio conversion relationship, a dictionary lookup mechanism based on business scale, and a rule engine;

[0116] Among them, the attribute information includes capacity and token generation speed;

[0117] The token generation speed is the speed at which tokens are generated and placed in the dynamic token bucket, and the capacity is the number of tokens that the dynamic token bucket can carry.

[0118] Optionally, the equity distribution device further includes: a model training module, which is used to:

[0119] Before obtaining the current marketing activity category, the current user portrait category, and the user behavior data, obtaining at least one historical marketing activity category, the historical user portrait category corresponding to each historical marketing activity category, and the historical user behavior data and historical traffic data corresponding to each historical user portrait category;

[0120] According to the correspondence between each historical marketing activity category, each historical user portrait category, each historical user behavior data and each historical traffic data, training samples with historical traffic data as labels are constructed respectively;

[0121] The initial traffic prediction model is trained through each training sample to generate a pre-trained traffic prediction model.

[0122] Optionally, the equity distribution device further includes: a supplementary training module for:

[0123] After the rights are distributed through the dynamic token bucket, user feedback data is obtained, and supplementary training samples are constructed based on the current marketing activity category, current user portrait category, user behavior data and user feedback data; the traffic prediction model is supplemented with training using the supplementary training samples.

[0124] Optionally, the pre-trained traffic prediction model is a multi-factor prediction model that uses both long short-term memory network units and recurrent neural network units;

[0125] The architecture of the long short-term memory network unit and the recurrent neural network unit in the pre-trained traffic prediction model includes at least one of the following:

[0126] Hybrid architecture;

[0127] Input-output hybrid architecture;

[0128] Multi-layered architecture.

[0129] The rights distribution device provided in the embodiment of the present invention can execute the rights distribution method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0130] Embodiment 4

[0131] Figure 5 A schematic diagram of the structure of a rights distribution device provided for Embodiment 4 of the present invention. The rights distribution device 40 may be intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The rights distribution device 40 may also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or claimed herein.

[0132] like Figure 5As shown, the equity distribution device 40 includes at least one processor 41, and a memory connected to the at least one processor 41, such as a read-only memory (ROM) 42, a random access memory (RAM) 43, etc., wherein the memory stores a programmable program that can be executed by at least one processor, and the processor 41 can perform various appropriate actions and processes according to the programmable program stored in the read-only memory (ROM) 42 or the programmable program loaded from the storage unit 48 to the random access memory (RAM) 43. In RAM 43, various programs and data required for the operation of the equity distribution device 40 can also be stored. The processor 41, ROM 42 and RAM 43 are connected to each other through a bus 44. An input / output (I / O) interface 45 is also connected to the bus 44. Optionally, the processor can be an FPGA.

[0133] Multiple components in the equity distribution device 40 are connected to the I / O interface 45, including: an input unit 46, such as a keyboard, a mouse, etc.; an output unit 47, such as various types of displays, speakers, etc.; a storage unit 48, such as a disk, an optical disk, etc.; and a communication unit 49, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 49 allows the equity distribution device 40 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0134] The processor 41 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 41 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 41 executes the various methods and processes described above, such as the equity distribution method.

[0135] In some embodiments, the equity distribution method can be implemented as a programmable program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 48. In some embodiments, part or all of the programmable program can be loaded and / or installed on the equity distribution device 40 via the ROM 42 and / or the communication unit 49. When the programmable program is loaded into the RAM 43 and executed by the processor 41, one or more steps of the equity distribution method described above can be performed. Alternatively, in other embodiments, the processor 41 can be configured to perform the equity distribution method in any other suitable manner (e.g., by means of firmware).

[0136] Optionally, an embodiment of the present invention further provides a computer program product, including a computer program, which, when executed by a processor, implements the equity distribution method provided in any embodiment of the present invention.

[0137] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include: being implemented in one or more programmable programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0138] Programmable programs for implementing the method of the present invention may be written in any combination of one or more programming languages. These programmable programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the programmable program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The programmable program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0139] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a programmable program for use by or in conjunction with an instruction execution system, device, or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0140] To provide interaction with a user, the systems and techniques described herein may be implemented on a rights distribution device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the rights distribution device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0141] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0142] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by programmable programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0143] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0144] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for distributing rights and interests, characterized in that: include: Get the current marketing campaign category, current user profile category, and user behavior data; Inputting the current marketing activity category, the current user profile category and the user behavior data into a pre-trained traffic prediction model to determine a traffic prediction value; The attribute information of the dynamic token bucket is adjusted according to the traffic prediction value, and the equity is distributed through the dynamic token bucket.

2. The equity distribution method according to claim 1, characterized in that: The adjusting the attribute information of the dynamic token bucket according to the traffic prediction value includes: Determine attribute information of a dynamic token bucket according to the traffic prediction value and the pre-built traffic parameter correspondence; The pre-built traffic parameter correspondence relationship includes at least one of a proportional conversion relationship, a dictionary lookup mechanism based on business scale, and a rule engine; Wherein, the attribute information includes capacity and token generation speed; The token generation speed is the speed at which tokens are generated and placed in the dynamic token bucket, and the capacity is the number of tokens that the dynamic token bucket can carry.

3. The equity distribution method according to claim 2, characterized in that: The equity distribution through the dynamic token bucket includes: When a user's rights and interests request is received, a token is taken out from the dynamic token bucket, and rights and interests are distributed to the user through the token.

4. The equity distribution method according to claim 1, characterized in that: Before obtaining the current marketing campaign category, current user profile category and user behavior data, including: Obtain at least one historical marketing activity category, a historical user profile category corresponding to each of the historical marketing activity categories, and historical user behavior data and historical traffic data corresponding to each of the historical user profile categories; According to the correspondence between each of the historical marketing activity categories, each of the historical user portrait categories, each of the historical user behavior data and each of the historical traffic data, respectively construct training samples with the historical traffic data as labels; The initial traffic prediction model is trained by using the training samples to generate a pre-trained traffic prediction model.

5. The equity distribution method according to any one of claims 1 to 4, characterized in that: After the equity distribution is performed through the dynamic token bucket, the method further includes: Acquire user feedback data, and construct supplementary training samples according to the current marketing activity category, the current user portrait category, the user behavior data, and the user feedback data; The traffic prediction model is supplemented with training using the supplementary training samples.

6. The equity distribution method according to claim 5, characterized in that: The pre-trained traffic prediction model is a multi-factor prediction model that uses long short-term memory network units and recurrent neural network units at the same time; The architecture of the long short-term memory network unit and the recurrent neural network unit in the pre-trained traffic prediction model includes at least one of the following: Hybrid architecture; Input-output hybrid architecture; Multi-layered architecture.

7. A rights distribution device, characterized in that: include: Data acquisition module, used to obtain the current marketing activity category, current user portrait category and user behavior data; A traffic prediction module, used to input the current marketing activity category, the current user portrait category and the user behavior data into a pre-trained traffic prediction model to determine a traffic prediction value; The equity distribution module is used to adjust the attribute information of the dynamic token bucket according to the traffic prediction value, and distribute the equity through the dynamic token bucket.

8. A rights distribution device, characterized in that: include: at least one processor; and a memory communicatively coupled to the at least one processor; Wherein, the memory stores a programmable program that can be executed by the at least one processor, and the programmable program is executed by the at least one processor so that the at least one processor can execute the equity distribution method described in any one of claims 1-6.

9. A storage medium containing computer executable instructions, characterized in that: The executable instructions, when executed by a programmable processor, are used to execute the rights distribution method as described in any one of claims 1-6.

10. A computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the method for distributing rights and interests as claimed in any one of claims 1 to 6 is implemented.

Images